sh0wie/Qwen3.8-Flash-Next-MTP-Drafter-MLX-bf16
Same weights, repacked (fewer layers) vs Qwen/Qwen3.8-Flash-Next. sh0wie/Qwen3.8-Flash-Next-MTP-Drafter-MLX-bf16 matches part of Qwen/Qwen3.8-Flash-Next (29 of its 1658 tensors; the rest of Qwen/Qwen3.8-Flash-Next has no counterpart). On the shared part: Same weights in different files (re-saved, re-sharded or new metadata), plus 3 new tensors.
Tensors changed: Norms 0/8, other 0/3, Attention 0/8, MLP 0/8, Embeddings 0/2